Underwater acoustic tomography parallel computing method, system and equipment based on M sequence parallel demodulation and medium

Abstract

The invention relates to an underwater acoustic tomography parallel computing method, system and device based on M sequence parallel demodulation and a medium, and the method comprises the steps: executing an initialization operation, and loading configuration parameters; determining a parallel computing segment number according to the output data length, dividing the acoustic tomography acquisition orthogonal signal into a plurality of data segments with time sequence overlapping characteristics according to the parallel computing segment number, and distributing an independent parallel computing thread for each data segment; in each thread, generating a symbol coefficient based on an M sequence bit value, dynamically distributing a signal sample window in combination with a modulation expansion factor, and performing symbol weighting and accumulation operation by bit-by-bit iteration until all M sequence bits in a corresponding data segment are processed; and combining processing results output by all the threads to generate a final correlation sequence, and performing inversion based on the final correlation sequence to obtain underwater acoustic tomography information. Therefore, breakthrough improvement of the demodulation efficiency of the underwater acoustic tomography signal is realized through a staged parallelization processing architecture.

Description

Technical Field

[0001] This invention relates to the fields of underwater acoustic engineering and signal processing technology, and in particular to a parallel computing method, system, device, and medium for underwater acoustic tomography based on M-sequence parallel demodulation. Background Technology

[0002] Underwater acoustic tomography is an observational technique that uses acoustic signals to invert environmental parameters of aquatic bodies. Its basic principle is to reconstruct the distribution of internal parameters (such as temperature and flow velocity) of the medium by measuring the propagation time, amplitude attenuation, or phase change of sound waves in the medium, combined with an inverse problem-solving algorithm. In the implementation of this technique, demodulation is a crucial step in extracting effective acoustic signals from background noise; current mainstream schemes all employ the M-sequence cross-correlation algorithm to achieve signal separation.

[0003] Traditional M-sequence demodulation methods typically employ sliding correlators, which work by performing point-by-point sliding correlation calculations between the local M-sequence and the received signal. However, due to the large volume and long M-sequence codes of underwater acoustic tomography data, traditional serial demodulation methods are extremely time-consuming, severely limiting data processing efficiency and making it difficult to meet the real-time requirements of applications (such as rapid marine environmental data acquisition and inversion). While the principle of M-sequence demodulation is mature, there is a lack of efficient and dedicated demodulation schemes for large-scale underwater acoustic tomography data. Therefore, there is an urgent need for a new method that can significantly improve demodulation speed while maintaining demodulation accuracy.

[0004] It is worth noting that current demodulation methods are designed based on serial computing architectures, which are difficult to meet the requirements of real-time performance and high-resolution reconstruction. Especially when transmitting high-resolution, large-scale water body observation data, the capacity of the transmitted data far exceeds the demodulation processing speed, falling far short of the practical application indicators of real-time performance and high resolution. Summary of the Invention

[0005] (a) Technical problems to be solved In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a parallel computing method, system, device and medium for underwater acoustic tomography based on M-sequence parallel demodulation. It solves the technical problem that the existing serial demodulation method based on M-sequence cross-correlation algorithm has a surge in computing time due to the large scale of underwater acoustic tomography data and the significant increase in code length, resulting in a serious imbalance between data processing speed and high-resolution observation data transmission rate, which cannot meet the application requirements of real-time environmental parameter inversion and high-precision reconstruction.

[0006] (II) Technical Solution To achieve the above objectives, the main technical solutions adopted by the present invention include: In a first aspect, embodiments of the present invention provide a parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation, comprising: In response to external commands or conditional triggers, perform initialization operations and load configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length. The number of parallel computing segments is determined based on the length of the output data. The orthogonal signals acquired by acoustic tomography are divided into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and an independent parallel computing thread is assigned to each data segment. In each thread, symbol coefficients are generated based on the bit values ​​of the M sequence. The signal sample window is dynamically allocated in combination with the modulation spread factor. Symbol weighting and accumulation operations are performed bit by bit until all M sequence bits in the corresponding data segment are processed. The processing results of all threads are merged to generate the final correlation sequence, and underwater acoustic tomography information is obtained by inversion based on the final correlation sequence.

[0007] Optionally, in response to external commands or conditional triggers, an initialization operation is performed, loading configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length parameters, including: In response to external commands or conditions, perform initialization operations; Load the M sequence in binary form; The orthogonal input signal required for acoustic tomography acquisition is loaded with external input, wherein the orthogonal input signal contains discrete sampling sequences of sine and cosine waves; An externally input modulation spread factor is loaded to regulate the mapping relationship between the M-sequence bits and the number of signal samples; Load the output data length parameter and pre-allocate memory space for the output storage array based on the output data length parameter.

[0008] Optionally, the number of parallel computing segments is determined based on the output data length. The orthogonal signals acquired by acoustic tomography are divided into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and an independent parallel computing thread is allocated to each data segment, including: The number of parallel computing segments is determined based on the output data length parameter, so that the number of parallel computing segments is equal to the output data dimension. Based on the preset sliding step size parameter, starting from the initial time point of the quadrature input signal, the data segments are divided sequentially, and the starting position of each subsequent data segment is progressively increased by the sliding step size relative to the starting position of the previous segment. Multiple data segments with temporal overlap characteristics are generated by progressive partitioning. The length of the overlapping area between adjacent data segments is the difference between the sliding step size and the data segment length. Each partitioned data segment is assigned an independent thread, and a multi-threaded parallel computing architecture is deployed to enable each thread to independently handle the computing tasks of its corresponding data segment.

[0009] Optionally, in each thread, symbol coefficients are generated based on the M-sequence bit values. The signal sample window is dynamically allocated in conjunction with the modulation spread factor, and symbol weighting and accumulation operations are performed bit by bit until all M-sequence bits within the corresponding data segment are processed, including: For each data segment allocated to a parallel computing thread, the initial values ​​of the preset sine accumulator and cosine accumulator are set to zero, and the starting position of the signal pointer is located according to the starting position of the data segment. Read the current bit value sequentially according to the binary encoding order of the M sequence bits, map the current bit value as logic 1 to a positive sign coefficient, and logic 0 to a negative sign coefficient, and generate a sign weighting parameter that corresponds one-to-one with the M sequence bits; Based on the modulation spread factor, the signal sample window corresponding to each M sequence bit is determined, and continuous signal samples within the signal sample window are read from the starting position of the signal pointer; The continuous signal samples are weighted according to the symbol weighting parameters, and the weighted sine component sample values ​​are accumulated into the sine accumulator, and the cosine component sample values ​​are accumulated into the cosine accumulator. After completing the accumulation operation of the current M-sequence bit, move the signal pointer to the next processing start position according to the length of the signal sample window, and iterate through the next M-sequence bit until all M-sequence bits in the current data segment have been processed.

[0010] Optionally, performing a weighting operation on continuous signal samples according to the sign weighting parameter and accumulating the weighted sine component sample values ​​into a sine accumulator, and accumulating the cosine component sample values ​​into a cosine accumulator includes: From the current signal pointer position, synchronously read the sine component sample value and the cosine component sample value in the signal sample window to form an orthogonal signal pair; The sign weighting parameters are simultaneously injected into the preprocessing channels of the sine component samples and the cosine component samples to generate a sign-aligned orthogonal component sequence. The sign-weighted sine component samples are sequentially superimposed onto the sine accumulator in chronological order to complete the accumulation of all sine components within the current window; The sign-weighted cosine component samples are superimposed one by one onto the cosine accumulator in the same time order to complete the accumulation of all cosine components in the current window; After each sample accumulation, the update count of the sine accumulator and cosine accumulator is simultaneously detected to ensure that the total number of accumulated samples is consistent and equal to the window length.

[0011] Optionally, merging the processing results from all threads to generate the final correlation sequence includes: The final values ​​of the sine accumulator and cosine accumulator of each thread are stored in independent memory areas to form segmented values ​​that correspond one-to-one with the data segments. Repeatedly perform the weighted accumulation operation, dynamically update the symbol weighting parameters in each iteration, and generate updated segmented values ​​that are bound to the iteration cycle; All thread segment correlation values ​​are spatiotemporally aligned according to the acoustic tomography channel number to generate the final correlation sequence.

[0012] Optionally, underwater acoustic tomography information obtained based on the final correlation sequence inversion includes: Based on the final correlation sequence, the time delay distribution characteristics of the sound wave are located by searching for correlation extreme points in the sequence, and the amplitude values ​​of the corresponding extreme points are extracted as amplitude attenuation features to characterize the propagation characteristics of the sound wave in the target water area. Based on the propagation characteristics of sound waves, the time delay distribution characteristics and amplitude attenuation characteristics are mapped to spatial position parameters in the vertical profile direction of the target water area; Based on spatial location parameters, a sound velocity gradient change profile and a medium density distribution map are generated, forming a quantifiable underwater acoustic tomography result.

[0013] Secondly, embodiments of the present invention provide an underwater acoustic tomography parallel computing system based on M-sequence parallel demodulation, comprising: The startup configuration module is used to perform initialization operations in response to external commands or conditional triggers, loading configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length. The thread allocation module is used to determine the number of parallel computing segments based on the length of the output data, divide the orthogonal signals acquired by acoustic tomography into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and allocate an independent parallel computing thread to each data segment. The parallel computing module is used to generate symbol coefficients based on the bit values ​​of the M sequence in each thread, dynamically allocate the signal sample window in combination with the modulation spread factor, and perform symbol weighting and accumulation operations bit by bit until all M sequence bits in the corresponding data segment are processed. The merging and inversion module is used to merge the processing results output by all threads to generate the final correlation sequence, and to invert the underwater acoustic tomography information based on the final correlation sequence.

[0014] Thirdly, embodiments of the present invention provide an underwater acoustic tomography parallel computing device based on M-sequence parallel demodulation, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation as described above.

[0015] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation as described above.

[0016] (III) Beneficial Effects The beneficial effects of this invention are: this invention achieves a breakthrough improvement in the demodulation efficiency of underwater acoustic tomography signals through a staged parallel processing architecture.

[0017] First, in response to external commands or conditional triggers, the demodulation process can be quickly and parameterized by loading preset M-sequences, orthogonal signals, modulation spread factors, and output length parameters, providing a dynamic adaptation basis for acoustic tomography observation tasks of different scales.

[0018] Secondly, the number of parallel computing segments with temporal overlap characteristics is dynamically divided based on the length of the output data. This not only ensures the independence of parallel threads, but also offsets the segment boundary effect through the overlap design, maintains the integrity of the demodulation results, and, combined with the multi-thread allocation mechanism, reconstructs the serial data stream into parallel executable units, so that the computing load is evenly distributed to the multi-core processor, significantly improving the utilization of hardware resources.

[0019] Next, within each thread, the parallelized symbol weighted accumulation is achieved through the generation of symbol coefficients driven by the M-sequence bit value and the dynamic sample window allocation controlled by the modulation spread factor. This avoids the sequential dependency bottleneck of the traditional sliding correlation algorithm and reduces the processing time of a single data segment to 1 / N of the original serial scheme (where N is the number of processor cores).

[0020] Furthermore, by merging the segmented correlation values ​​output by multiple threads and generating the final correlation sequence of spatiotemporal correlation, the distributed computing results are integrated while maintaining the spatiotemporal continuity of the signal. This ensures that the accuracy and real-time performance of acoustic tomography parameter inversion are improved simultaneously, completely solving the problem of imbalance between data processing rate and observation feedback requirements caused by excessive computation time in traditional serial demodulation methods, and meeting the timeliness requirements of high-resolution environment reconstruction. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the overall steps of the method provided in the embodiments of the present invention; Figure 2 A schematic diagram of the process framework of the method provided in the embodiments of the present invention; Figure 3 This is a schematic diagram illustrating the specific process of step S1 of the method provided in this embodiment of the invention; Figure 4 This is a detailed flowchart illustrating step S2 of the method provided in this embodiment of the invention; Figure 5 This is a detailed flowchart illustrating step S3 of the method provided in this embodiment of the invention; Figure 6 This is a detailed flowchart illustrating step S31 of the method provided in this embodiment of the invention; Figure 7 This is a detailed flowchart illustrating step S4 of the method provided in this embodiment of the invention; Figure 8 A set of comparison charts showing the actual processing of underwater data provided in the embodiments of the present invention; Figure 9 Another set of underwater data processing comparison diagrams provided in the embodiments of the present invention. Detailed Implementation

[0022] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] like Figure 1 As shown in the embodiment of the present invention, an underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation is proposed, comprising: responding to external commands or condition triggers, performing an initialization operation, loading configuration parameters including M-sequence, acoustic tomography acquisition orthogonal signals, modulation spread factor, and output data length parameters; determining the number of parallel computing segments according to the output data length, dividing the acoustic tomography acquisition orthogonal signals into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and allocating an independent parallel computing thread for each data segment; in each thread, generating symbol coefficients based on the M-sequence bit values, dynamically allocating signal sample windows in combination with the modulation spread factor, and performing symbol weighting and accumulation operations bit by bit until all M-sequence bit processing in the corresponding data segment is completed; merging the processing results output by all threads to generate a final correlation sequence, and inverting underwater acoustic tomography information based on the final correlation sequence.

[0024] This invention achieves a breakthrough improvement in the demodulation efficiency of underwater acoustic tomography signals through a staged parallel processing architecture.

[0025] First, in response to external commands or conditional triggers, the demodulation process can be quickly deployed by loading preset M-sequences, orthogonal signals, modulation spread factors and output length parameters, providing a dynamic adaptation basis for adapting to acoustic tomography observation tasks of different scales. Secondly, the number of parallel computing segments with temporal overlap characteristics is dynamically divided based on the length of the output data. This not only ensures the independence of parallel threads, but also offsets the segment boundary effect through the overlap design, maintains the integrity of the demodulation results, and, combined with the multi-thread allocation mechanism, reconstructs the serial data stream into parallel executable units, so that the computing load is evenly distributed to the multi-core processor, significantly improving the utilization of hardware resources.

[0026] Next, within each thread, the parallelized symbol weighted accumulation is achieved through the generation of symbol coefficients driven by the M-sequence bit value and the dynamic sample window allocation controlled by the modulation spread factor. This avoids the sequential dependency bottleneck of the traditional sliding correlation algorithm and reduces the processing time of a single data segment to 1 / N of the original serial scheme (where N is the number of processor cores).

[0027] Furthermore, by merging the segmented correlation values ​​output by multiple threads and generating the final correlation sequence of spatiotemporal correlation, the distributed computing results are integrated while maintaining the spatiotemporal continuity of the signal. This ensures that the accuracy and real-time performance of acoustic tomography parameter inversion are improved simultaneously, completely solving the problem of imbalance between data processing rate and observation feedback requirements caused by excessive computation time in traditional serial demodulation methods, and meeting the timeliness requirements of high-resolution environment reconstruction.

[0028] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0029] Specifically, refer to Figure 1 and Figure 2 This invention provides a parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation, comprising: S1. In response to external commands or conditional triggering, perform initialization operations and load configuration parameters including M-sequence, orthogonal signals acquired by acoustic tomography, modulation spread factor, and output data length parameters.

[0030] Furthermore, such as Figure 3 As shown, step S1 includes: S11. Execute initialization operations in response to external commands or conditions. Specifically, this refers to initiating the system initialization process through user-input control commands (such as program start commands) or preset condition signals (such as sensor threshold triggers, timed task activation, network protocol command reception, etc.).

[0031] S12. Load the M sequence in binary form. The M sequence is a finite-length encoded sequence composed of binary elements 0 and 1 arranged in a specific order. It is used to control the weighting and demodulation process of the signal through its bit value logic.

[0032] S13. Load the external input quadrature input signal required for acoustic tomography acquisition, wherein the quadrature input signal contains discrete sampling sequences of sine and cosine waves.

[0033] S14. Load the modulation spread factor from the external input to regulate the mapping relationship between the M-sequence bits and the number of signal samples.

[0034] S15. Load the output data length parameter and pre-allocate memory space for the output storage array based on the output data length parameter.

[0035] In one embodiment, the initialization phase first defines the core input parameters and completes memory pre-allocation, specifically including: (1) Binary M-sequence: M=[m1,m2,...,m L ], m j ∈{0,1}, the length of the M sequence is L.

[0036] (2) Orthogonal input signals: discrete sampling sequences of sine wave S(t)=sin(2πft) and cosine wave C(t)=cos(2πft), where f is the sound wave frequency.

[0037] (3) Modulation spreading factor: N mod This is used to control the number of signal samples corresponding to each M-sequence bit.

[0038] (4) Output data length: D, whose value directly determines the number of segments for parallel computing.

[0039] Based on this, the sine accumulation result array and the cosine accumulation result array are generated and initialized respectively through memory pre-allocation: Rsin=[0] 1×D , Rcos=[0] 1×D This provides a data container for subsequent parallel computing.

[0040] S2. Determine the number of parallel computing segments based on the length of the output data. Divide the orthogonal signals acquired by acoustic tomography into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and allocate an independent parallel computing thread to each data segment.

[0041] Furthermore, such as Figure 4 As shown, step S2 includes: S21. Determine the number of parallel computing segments based on the output data length parameter, so that the number of parallel computing segments is equal to the output data dimension.

[0042] S22. Based on the preset sliding step size parameter, starting from the initial time point of the quadrature input signal, the data segments are divided sequentially, and the starting position of each subsequent data segment is progressively increased by the sliding step size relative to the starting position of the previous segment.

[0043] S23. Multiple data segments with temporal overlap characteristics are generated through progressive partitioning. The length of the overlapping area between adjacent data segments is the difference between the sliding step size and the length of the data segment.

[0044] S24. Assign an independent thread to each divided data segment and deploy a multi-threaded parallel computing architecture so that each thread can independently handle the computing tasks of the corresponding data segment.

[0045] In one specific embodiment, when deploying a parallel computing architecture, the number of parallel computing segments, D, is first determined based on the output data length parameter, ensuring that it is strictly equal to the output data dimension. Based on this, the input signal segmentation process is specifically implemented as follows: starting from the initial time point of the orthogonal input signal, the data segments are sequentially divided based on a preset sliding step size parameter, such that the starting position of the (i+1)th segment satisfies t. i+1 =t i +Δt (where Δt≥1 is the sliding step size), thereby generating D data segments with temporal overlap characteristics.

[0046] According to the above rules, the length of the overlapping region between adjacent data segments is determined by the difference between the total length of the data segments and the sliding step size. By constraining Δt < data segment length, the continuity correction function of the overlapping region is ensured to be effective. Finally, an independent thread is assigned to each divided data segment, and a data-level parallel architecture drives all threads to simultaneously execute the computation tasks of the corresponding data segments, achieving efficient demodulation of large-scale underwater acoustic signals.

[0047] S3. In each thread, symbol coefficients are generated based on the M-sequence bit values. The signal sample window is dynamically allocated in combination with the modulation spread factor. Symbol weighting and accumulation operations are performed bit by bit until all M-sequence bits in the corresponding data segment are processed.

[0048] Furthermore, such as Figure 5 As shown, step S3 includes: S31. For each data segment allocated to a parallel computing thread, set the initial values ​​of the preset sine accumulator and cosine accumulator to zero, and locate the starting position of the signal pointer according to the starting position of the data segment.

[0049] S32. Read the current bit value sequentially according to the binary encoding order of the M sequence bits, map the current bit value as logic 1 to a positive sign coefficient, and logic 0 to a negative sign coefficient, and generate a sign weighting parameter that corresponds one-to-one with the M sequence bits.

[0050] S33. Based on the modulation spread factor, determine the signal sample window corresponding to each M sequence bit, and read the continuous signal samples in the signal sample window from the starting position of the signal pointer.

[0051] S34. Perform a weighting operation on the continuous signal samples according to the symbol weighting parameters, and accumulate the weighted sine component sample values ​​to the sine accumulator and the cosine component sample values ​​to the cosine accumulator.

[0052] Furthermore, such as Figure 6 As shown, step S34 includes: S341. From the current signal pointer position, synchronously read the sine component sample value and cosine component sample value in the signal sample window to form an orthogonal signal pair.

[0053] S342. Simultaneously inject the symbol weighting parameters into the preprocessing channels of the sine component samples and the cosine component samples to generate a symbol-aligned orthogonal component sequence.

[0054] S343. The symbol-weighted sine component samples are sequentially superimposed onto the sine accumulator in chronological order to complete the accumulation of all sine components in the current window.

[0055] S344. The symbol-weighted cosine component samples are superimposed onto the cosine accumulator one by one in the same time order to complete the accumulation of all cosine components in the current window.

[0056] S345. After each sample accumulation, synchronously detect the update count of the sine accumulator and the cosine accumulator to ensure that the total number of accumulated samples is consistent and equal to the window length.

[0057] S35. After completing the accumulation operation of the current M sequence bit, move the signal pointer to the next processing start position according to the length of the signal sample window, and iterate through the next M sequence bit until all M sequence bits in the current data segment have been processed.

[0058] refer to Figure 2 In one specific embodiment, the implementation steps of the parallel correlation calculation and physical signal accumulation algorithm based on the M-sequence are as follows: For the i-th data segment (1≤i≤D), first complete the initialization operation, and then set the sine component accumulator... Sum and cosine component accumulator All values ​​are cleared to zero, and the initial position of the signal pointer for the current data segment is set to p. i =t iThen iterate through each binary bit m of the M sequence. j (1≤j≤L), determine the sign based on the logical value of the current M sequence bit: if m j =1, then the sign coefficient α = +1; if m j If α = 0, then the sign coefficient α = -1.

[0059] Next, based on the determined sign coefficients, the continuous 2N values ​​in the input signal are processed. mod Each sample undergoes weighted summation to perform the following signal modulation: ; ; In the formula, cor sin The result of the cross-correlation accumulation of the sinusoidal carrier signal (Sin) and the M-sequence is cor. cos S is the result of the cross-correlation accumulation of the cosine carrier signal (Cos) and the M sequence. sin (i+m-1) is the discrete sampling sequence of a sinusoidal carrier, S cos (i+m-1) is the discrete sampling sequence of the cosine carrier.

[0060] Finally, after each calculation, slide the signal pointer position forward by 2N. mod Each sample unit is updated to p. i ←p i +2N mod This is to adapt to the processing requirements of the next M-sequence bits.

[0061] S4. Merge the processing results of all threads to generate the final correlation sequence, and invert the underwater acoustic tomography information based on the final correlation sequence.

[0062] Furthermore, such as Figure 7 As shown, step S4 includes: S41. Store the final values ​​of the sine accumulator and cosine accumulator of each thread into independent memory areas to form segmented values ​​that correspond one-to-one with the data segments.

[0063] S42. Repeatedly perform the weighted accumulation operation multiple times, dynamically update the symbol weighting parameters in each iteration, and generate updated segmented related values ​​bound to the iteration cycle.

[0064] S43. All thread segment correlation values ​​are spatiotemporally aligned according to the acoustic tomography channel number to generate the final correlation sequence.

[0065] S44. Based on the final correlation sequence, the time delay distribution characteristics of the sound wave are located by searching for correlation extrema in the sequence, and the amplitude values ​​of the corresponding extrema are extracted as amplitude attenuation features to characterize the propagation characteristics of the sound wave in the target water area. In this step, the correlation sequence data points are traversed, and the time delay feature points corresponding to the sound wave propagation path are located by detecting local maxima greater than the data values ​​of the K adjacent sampling points (K is a preset threshold, typically ≥3).

[0066] S45. Based on the propagation characteristics of sound waves, the time delay distribution characteristics and amplitude attenuation characteristics are mapped to spatial position parameters in the vertical profile direction of the target water area; wherein the spatial position parameters are the set data of vertical depth coordinates in the target water area and the sound velocity value and medium density value at that depth.

[0067] S46. Based on spatial location parameters, generate a sound velocity gradient profile and a medium density distribution map to form a quantifiable underwater acoustic tomography result. Specifically, in the sound velocity profile drawing, the sound velocity values ​​are arranged in ascending order of depth coordinates, and a continuous sound velocity gradient curve is generated through linear interpolation, with abrupt inflection points marked. Second, in the density distribution, discrete density values ​​are converted into contour maps within the vertical profile space, and chromaticity mapping characterizes density strength. Finally, the sound velocity gradient curve and density contour map are fused to form a vector-format underwater acoustic tomography map, with the coordinate axes labeled with the actual physical scale.

[0068] In another embodiment, for each data segment after parallel computation is completed, the sine accumulator value of the i-th thread is first... With cosine accumulator value Each segment is stored in a separate memory address space, forming a segmented value that strictly corresponds to the data segment. .

[0069] During this process, a weighted accumulation operation with dynamically updated symbol coefficients is performed through multiple iterations to ensure that the segmented correlation values ​​generated in each iteration are bound to the physical state of the current parameter configuration period. Finally, the segmented correlation values ​​of all threads are spatiotemporally aligned and merged according to the acoustic tomography channel number, outputting the complete correlation sequence as follows: .

[0070] The key expression for the signal modulation matching sine operation is as follows: ; Where, α j =2m j -1∈{-1,+1}, cosine is similar to sine.

[0071] Therefore, for reference Figure 8 and Figure 9 Two sets of underwater data processing comparison experiment results, among which Figure 8Based on the Ryzen R5 5500 6-core 12-processor processor platform, Figure 9 The system utilizes an Intel i5 13400f 10-core, 16-processor platform. Specific computational efficiency can be quantified using the statistical data in Table 1: When processing 10,000 data samples, the theoretical performance improvement is 6 times with a 6-core processor configuration, while the actual measured improvement is 5.98 times. When expanded to an 8-core configuration, the theoretical improvement increases to 8 times, while the actual efficiency reaches 7.87 times. This result demonstrates that the actual performance of multi-core parallel computing highly matches the theoretical expectation, verifying the efficiency gain characteristics of the proposed solution.

[0072] Table 1 Calculation Efficiency Statistics

[0073] Additionally, this embodiment of the invention provides an underwater acoustic tomography parallel computing system based on M-sequence parallel demodulation, comprising: a startup configuration module, used to perform initialization operations in response to external commands or condition triggers, loading configuration parameters including M-sequence, acoustic tomography acquisition orthogonal signals, modulation spread factor, and output data length parameters; a thread allocation module, used to determine the number of parallel computing segments according to the output data length, divide the acoustic tomography acquisition orthogonal signals into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and allocate an independent parallel computing thread to each data segment; a parallel computing module, used to generate symbol coefficients based on the M-sequence bit values ​​in each thread, dynamically allocate signal sample windows in combination with the modulation spread factor, and perform symbol weighting and accumulation operations bit by bit until all M-sequence bit processing in the corresponding data segment is completed; and a merging and inversion module, used to merge the processing results output by all threads to generate a final correlation sequence, and invert underwater acoustic tomography information based on the final correlation sequence.

[0074] Furthermore, this embodiment of the invention also provides a parallel computing device for underwater acoustic tomography based on M-sequence parallel demodulation, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described above. Through this hardware and software co-design, the processor can dynamically schedule multi-core computing resources and precisely control the entire computational process of M-sequence demodulation, signal segmentation, parallel weighted accumulation, and result synthesis.

[0075] Furthermore, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions. When executed by a processor, these instructions implement the underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation described above. When the medium is loaded into a computing device, the instruction code stored within it is parsed and executed line by line by the processor, thereby completely reproducing all the technical steps of the parallel computing method, including but not limited to dynamic parameter configuration, multi-threaded data partitioning, M-sequence parallel related operations, and physical information fusion output. The medium solidifies the technical solution through standardized coding, ensuring the portability and execution consistency of the method across different hardware platforms.

[0076] In summary, the embodiments of the present invention provide a parallel computing method, system, device, and medium for underwater acoustic tomography based on M-sequence parallel demodulation, and reference... Figure 2 The overall logic can be summarized as follows: Step 1, initialization and physical signal parameter configuration, configuring the M-sequence binary array, sequence length, acoustic tomography acquisition orthogonal signals, modulation spread factor, and pre-allocated memory; Step 2, parallel computing architecture deployment and underwater data segment division, dividing the dataset into N underwater signal segments, with the segment number being the output data length, and adopting a multi-threaded parallel processing mechanism, with each thread handling the computation task of one data segment; Step 3, parallel correlation calculation and physical signal accumulation based on the M-sequence, performing positive / negative correlation accumulation on continuous signal samples according to the M-sequence bit values, and controlling the number of signal samples corresponding to each M-sequence bit through the modulation spread factor; Step 4, result merging and physical information extraction.

[0077] Compared to existing acoustic tomography signal modulation algorithms, this invention improves the computational efficiency of acoustic tomography signal demodulation by a factor of n, where n is the number of processor cores, through a dynamic modulation mechanism and parallel optimization architecture. Specifically, the solution of this invention significantly improves computational efficiency and hardware resource utilization through parallel structure design, enabling underwater acoustic tomography signal processing to achieve real-time and high-resolution application indicators in terms of computational speed.

[0078] In terms of demodulation efficiency, by reconstructing the serial sliding correlation process into a parallel multi-channel correlation process, the computing power of modern multi-core processors is fully utilized. The demodulation time can theoretically be shortened to about 1 / N of the original time (where N is the number of processor cores), effectively solving the timeliness problem of underwater acoustic signal demodulation under large data volumes.

[0079] In terms of performance advantages, firstly, this invention parallelizes the algorithm at the algorithm level without changing the core mathematical principles of M-sequence demodulation, thus fully maintaining the demodulation accuracy and reliability of traditional methods. Secondly, this invention is not only applicable to marine acoustic tomography but can also be widely applied to other fields using M-sequences or similar spreading codes, such as underwater acoustic communication, underwater acoustic positioning, and sonar. Thirdly, based on advanced parallel programming tools such as MATLAB's parfor, the implementation of this invention is simple and efficient, reducing the complexity of engineering implementation and facilitating promotion and application. These technical characteristics collectively construct an advanced underwater acoustic signal processing solution that balances efficiency, accuracy, and universality.

[0080] Since the systems / devices described in the above embodiments of the present invention are systems / devices used to implement the methods of the above embodiments of the present invention, those skilled in the art can understand the specific structure and modifications of the systems / devices based on the methods described in the above embodiments of the present invention, and therefore will not be repeated here. All systems / devices used in the methods of the above embodiments of the present invention fall within the scope of protection of the present invention.

[0081] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0082] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions.

[0083] It should be noted that any reference numerals placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims that enumerate several means, several of these means may be embodied by the same hardware. The use of the terms first, second, third, etc., is merely for convenience of expression and does not indicate any order. These terms can be understood as part of the component names.

[0084] Furthermore, it should be noted that in the description of this specification, the terms "one embodiment," "some embodiments," "embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0085] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the claims should be interpreted to include both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0086] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, then this invention should also include these modifications and variations.

Claims

1. A parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation, characterized in that, include: In response to external commands or conditional triggers, perform initialization operations and load configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length. The number of parallel computing segments is determined based on the length of the output data. The orthogonal signals acquired by acoustic tomography are divided into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and an independent parallel computing thread is assigned to each data segment. In each thread, symbol coefficients are generated based on the bit values ​​of the M sequence. The signal sample window is dynamically allocated in combination with the modulation spread factor. Symbol weighting and accumulation operations are performed bit by bit until all M sequence bits in the corresponding data segment are processed. The processing results of all threads are merged to generate the final correlation sequence, and underwater acoustic tomography information is obtained by inversion based on the final correlation sequence.

2. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in claim 1, characterized in that, In response to external commands or conditional triggers, an initialization operation is performed, loading configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length. In response to external commands or conditions, perform initialization operations; Load the M sequence in binary form; The orthogonal input signal required for acoustic tomography acquisition is loaded with external input, wherein the orthogonal input signal contains discrete sampling sequences of sine and cosine waves; An externally input modulation spread factor is loaded to regulate the mapping relationship between the M-sequence bits and the number of signal samples; Load the output data length parameter and pre-allocate memory space for the output storage array based on the output data length parameter.

3. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in claim 1, characterized in that, The number of parallel computing segments is determined based on the output data length. The orthogonal signals acquired by acoustic tomography are divided into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and an independent parallel computing thread is assigned to each data segment, including: The number of parallel computing segments is determined based on the output data length parameter, so that the number of parallel computing segments is equal to the output data dimension. Based on the preset sliding step size parameter, starting from the initial time point of the quadrature input signal, the data segments are divided sequentially, and the starting position of each subsequent data segment is progressively increased by the sliding step size relative to the starting position of the previous segment. Multiple data segments with temporal overlap characteristics are generated by progressive partitioning. The length of the overlapping area between adjacent data segments is the difference between the sliding step size and the data segment length. Each partitioned data segment is assigned an independent thread, and a multi-threaded parallel computing architecture is deployed to enable each thread to independently handle the computing tasks of its corresponding data segment.

4. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in claim 1, characterized in that, In each thread, symbol coefficients are generated based on the M-sequence bit values. The signal sample window is dynamically allocated using the modulation spread factor, and symbol weighting and accumulation operations are performed bit-by-bit iteratively until all M-sequence bits within the corresponding data segment are processed, including: For each data segment allocated to a parallel computing thread, the initial values ​​of the preset sine accumulator and cosine accumulator are set to zero, and the starting position of the signal pointer is located according to the starting position of the data segment. Read the current bit value sequentially according to the binary encoding order of the M sequence bits, map the current bit value as logic 1 to positive sign coefficient, and logic 0 to negative sign coefficient, and generate sign weighting parameters that correspond one-to-one with the M sequence bits; Based on the modulation spread factor, the signal sample window corresponding to each M sequence bit is determined, and continuous signal samples within the signal sample window are read from the starting position of the signal pointer; The continuous signal samples are weighted according to the symbol weighting parameters, and the weighted sine component sample values ​​are accumulated into the sine accumulator, and the cosine component sample values ​​are accumulated into the cosine accumulator. After completing the accumulation operation of the current M-sequence bit, move the signal pointer to the next processing start position according to the length of the signal sample window, and iterate through the next M-sequence bit until all M-sequence bits in the current data segment have been processed.

5. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in claim 4, characterized in that, Performing a weighting operation on continuous signal samples according to the sign weighting parameter, and accumulating the weighted sine component sample values ​​into a sine accumulator, and accumulating the cosine component sample values ​​into a cosine accumulator includes: From the current signal pointer position, synchronously read the sine component sample value and the cosine component sample value in the signal sample window to form an orthogonal signal pair; The sign weighting parameters are simultaneously injected into the preprocessing channels of the sine component samples and the cosine component samples to generate a sign-aligned orthogonal component sequence. The sign-weighted sine component samples are sequentially superimposed onto the sine accumulator in chronological order to complete the accumulation of all sine components within the current window; The sign-weighted cosine component samples are superimposed one by one onto the cosine accumulator in the same time order to complete the accumulation of all cosine components in the current window; After each sample accumulation, the update count of the sine accumulator and cosine accumulator is simultaneously detected to ensure that the total number of accumulated samples is consistent and equal to the window length.

6. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in claim 4, characterized in that, The processing results from merging all threads' outputs to generate the final correlation sequence include: The final values ​​of the sine accumulator and cosine accumulator of each thread are stored in independent memory areas to form segmented values ​​that correspond one-to-one with the data segments. Repeatedly perform the weighted accumulation operation, dynamically update the symbol weighting parameters in each iteration, and generate updated segmented values ​​that are bound to the iteration cycle; All thread segment correlation values ​​are spatiotemporally aligned according to the acoustic tomography channel number to generate the final correlation sequence.

7. The parallel computing method for underwater acoustic tomography based on M-sequence parallel demodulation as described in any one of claims 1-6, characterized in that, The underwater acoustic tomography information obtained based on the final correlation sequence inversion includes: Based on the final correlation sequence, the time delay distribution characteristics of the sound wave are located by searching for correlation extreme points in the sequence, and the amplitude values ​​of the corresponding extreme points are extracted as amplitude attenuation features to characterize the propagation characteristics of the sound wave in the target water area. Based on the propagation characteristics of sound waves, the time delay distribution characteristics and amplitude attenuation characteristics are mapped to spatial position parameters in the vertical profile direction of the target water area; Based on spatial location parameters, a sound velocity gradient change profile and a medium density distribution map are generated, forming a quantifiable underwater acoustic tomography result.

8. A parallel computing system for underwater acoustic tomography based on M-sequence parallel demodulation, characterized in that, include: The startup configuration module is used to perform initialization operations in response to external commands or conditional triggers, loading configuration parameters including the M-sequence, the quadrature signal acquired by acoustic tomography, the modulation spread factor, and the output data length. The thread allocation module is used to determine the number of parallel computing segments based on the length of the output data, divide the orthogonal signals acquired by acoustic tomography into several data segments with temporal overlap characteristics according to the number of parallel computing segments, and allocate an independent parallel computing thread to each data segment. The parallel computing module is used to generate symbol coefficients based on the bit values ​​of the M sequence in each thread, dynamically allocate the signal sample window in combination with the modulation spread factor, and perform symbol weighting and accumulation operations bit by bit until all M sequence bits in the corresponding data segment are processed. The merging and inversion module is used to merge the processing results output by all threads to generate the final correlation sequence, and to invert the underwater acoustic tomography information based on the final correlation sequence.

9. A parallel computing device for underwater acoustic tomography based on M-sequence parallel demodulation, characterized in that, include: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which are executed by at least one processor to enable the at least one processor to perform the underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation as described in any one of claims 1-7.

10. A computer-readable storage medium storing computer-executable instructions thereon, characterized in that, When the executable instructions are executed by the processor, they implement the underwater acoustic tomography parallel computing method based on M-sequence parallel demodulation as described in any one of claims 1-7.

Images